Pith. sign in

REVIEW 5 major objections 5 minor 61 references

Early Detection of Multiwavelength Blazar Variability

T0 review · 5 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read An unsupervised deep-learning pipeline detects flaring states in multiwavelength blazar light curves without needing labeled flare examples.

desk verdict A useful, well-documented anomaly-detection framework with public code, but the headline claim of reliable historical detection is weakened by an in-sample evaluation that removes the ground-truth flares from the training background. read the letter →

arxiv 2411.10140 v2 pith:JYCLYEE7 submitted 2024-11-15 astro-ph.HE astro-ph.IM

Hermann Stolte , Jonas Sinapius , Iftach Sadeh , Elisa Pueschel , Matthias Weidlich , David Berge This is my paper · ORCID
classification astro-ph.HEastro-ph.IM
keywords blazaranomalydetectionmultiwavelengthlightcurvesrecurrentneuralnetworkvariationalautoencoderGaussianmixturemodelFermi-LATCTAO
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper presents a deep-learning framework for detecting flaring episodes in multiwavelength blazar light curves without labeled flares. It claims the framework can flag clear high states, downward deviations, and subtle correlations across bands, and that it detects known historical flares of BL Lacertae at the same time or earlier than the community's alert system. The significance output is a single calibrated number that propagates measurement uncertainties, designed for real-time use. A sympathetic reader would care because the method offers a standardized trigger for very-high-energy observations and a common language for comparing activity across different instruments.

What carries the argument

The core mechanism is an encoder-decoder recurrent neural network that forecasts background activity from a context window, followed by a variational autoencoder that maps the residuals (data minus forecast) into an embedding vector. A Bayesian Gaussian mixture model then scores how far an embedding lies from background, while the autoencoder's reconstruction error catches out-of-distribution states; the two p-values are combined into a single significance. Temporal weights suppress old anomalies in sparse channels, and uncertainty resampling propagates measurement errors into the final significance.

What would settle it

Take the BL Lacertae light curves, intentionally leave a weak flare (below the γsignoise = 5 cutoff) in the training data, retrain the pipeline, and check whether the calibrated significance for known historical flares decreases; a shift would show the background model is contaminated.

Watch

Extended reading notes

Core claim

The central claim is that an unsupervised anomaly-detection pipeline, trained only on cleaned background activity, can reliably flag flaring states in real-world multiwavelength light curves and can hint at precursors before the flare fully develops. On eleven years of BL Lacertae data spanning very-high-energy gamma rays to radio, the framework recovers five known VHE flares at multi-sigma significance, and in the 2019–2020 episodes it reached 3σ on the optical and gamma-ray precursor two to four days before the community's alerts. The detection is carried by two complementary test statistics: a mixture-model compatibility in a learned embedding space and a reconstruction-error statistic that catches states too novel for the embedding to represent.

Load-bearing premise

The framework defines 'anomaly' relative to a training sample that is assumed to be pure background, but that sample is produced by manually removing flaring intervals and applying per-channel variability cutoffs, so any hidden flares left in the background or any normal variability removed will bias the anomaly scores.

Editorial extensions

If this is right

  • The pipeline yields a single, calibrated significance per time step across all channels, so observatories can trigger VHE follow-up on a common numeric threshold.
  • Historical BL Lacertae analysis shows detections at the same time or earlier than the community's alert system, including a 3σ precursor before the 2020 optical and gamma-ray maximum.
  • The framework detects downward anomalies and cross-channel correlations, not just elevated fluxes, expanding the kinds of flaring states that can trigger alerts.
  • Measurement uncertainties propagate to the significance, so a low-quality outlier carries a large uncertainty and can be recognized as a likely false positive.
  • The method is built for real-time use and can be deployed in an automated alert broker without labeled flare data.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the background model generalizes, the same pipeline could be applied to other blazars or to transients like tidal disruption events, wherever a quiescent baseline can be defined.
  • The authors leave timescale flexibility implicit: an ensemble of pipelines with different time-bin widths could cover hour-to-week variability and might reveal precursors that day-binned analysis misses.
  • A direct test suggested by the approach is to run the framework on a second well-monitored blazar with known historical flares and compare detection times against the published alert record.
  • The definition of anomaly depends on the cleaned training sample, so the method's 'anomaly' is source-specific rather than universal; transferring it to a new source would require retraining on that source's own quiescent states.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper presents a deep-learning anomaly-detection framework for multiwavelength blazar light curves. The pipeline uses an RNN forecaster to predict background activity, an autoencoder to embed weighted residuals, and a Bayesian Gaussian mixture model to characterize background states; two test statistics (reconstruction error and mixture-model compatibility) are combined into a calibrated significance. The method is evaluated on simulations of the blazar 1ES 1215+303 in Fermi-LAT and CTAO bands and on historical multiwavelength data of BL Lacertae (VERITAS, Fermi-LAT, Swift-XRT, optical, SMA). The authors report that the framework detects known historical flares, sometimes earlier than the community alerts, and can hint at precursors.

Significance. If the central validation were out-of-sample, the framework would be a useful real-time tool for triggering IACT observations and for standardizing multiwavelength flare alerts. The paper has clear strengths: the code and simulation data are publicly available, measurement uncertainties are propagated through the pipeline, the simulations use realistic instrument responses, and the architecture is described in enough detail to be reproduced. However, the headline claim is currently supported only by an in-sample evaluation, so the significance of the result is conditional on a successful out-of-sample test.

major comments (5)
  1. [§4.1, §4.3, §4.4] The historical evaluation is in-sample with respect to the definition of ground truth. In §4.1 the training background is constructed by manually removing the VERITAS flaring intervals at MJDs 55,725, 57,200, 57,675, 58,100, 58,600 (plus 50-day buffers and per-channel cutoffs), and the high-significance peaks reported in §4.3 (t=2,500, 2,900, 3,400) and in Table 2 coincide with those same excluded intervals. A detector trained on the complement of the rule that flags those intervals will tend to flag them, so the reported detections and precursors may encode the cleaning thresholds rather than a learned multiwavelength precursor pattern. Please add a temporal out-of-sample test (e.g., train only on data before a cut date and evaluate after, or leave-one-flare-out) and report the false-alarm rate on quiescent periods; without such a control the headline claim in §4.4 is not established.
  2. [§3.4, §3.6] The simulation study is not an independent blind test. The light curves are generated from the 2017 steady-state spectrum of 1ES 1215+303, and the injected flares are scaled versions of the same Fref template (Valverde et al. 2020) used to define the source; the background is simulated from the same spectral model rather than cleaned real data. The observed scaling of significance with flare strength and coverage (Fig. 5) is a useful sensitivity check, but it cannot calibrate the historical comparison because there is no unknown signal and no realistic correlated multiwavelength background. Please either add a validation on a second source or on a held-out flare template/epoch, or explicitly restrict the simulation to a sensitivity study and base the detection claim solely on an out-of-sample real-data test.
  3. [Eq. (1)] Equation (1) is internally inconsistent with the intended behavior of the temporal weights. For tstep - tdelay > 1, the factor (1 - max(0, tstep - tdelay)) is negative; with gamma_decay = 1 this gives a negative weight, and with non-integer gamma_decay it gives a complex value, whereas the text and Fig. 2 describe weights that decay from unity toward zero. If this is a typo, please correct the sign (e.g., (1 + max(0, tstep - tdelay))^{-gamma_decay}) and confirm that the implementation matches; if the formula is correct, please explain how negative/complex weights are handled, since this equation controls the entire embedding input in §2.3.
  4. [§3.1, §3.2] The cleaning and augmentation procedure trains the background on data whose temporal ordering and inter-band correlations have been deliberately destroyed: data are shuffled inside each window, timestamps are randomized, and independent global shifts are applied per channel. The paper motivates the framework by sensitivity to 'subtle correlations across bands' (§1) and reports precursor signals that are inter-channel (§4.3.3), but no validation is shown that the correlation-free null distribution used for calibration matches real quiescent BL Lac activity. If real quiescent data contain residual correlations, the significance calibration will be systematically liberal. Please quantify this by comparing the background model distribution to un-cleaned quiescent data, or by demonstrating on simulations with correlated background that the false-alarm rate is controlled.
  5. [§4.3.3, Table 2] The 'same time or earlier than the state-of-the-art' claim is not supported by a defined detection threshold. Several entries in Table 2 are called detections or precursors at 1.3–2.5σ (e.g., t=3,434–3,435, t=570, t=3,940), while others are quoted at 3–10σ; the paper never states the significance threshold used to decide that an event was 'detected' by the framework, nor the expected number of false alarms over the 4,000-day baseline. Please specify the decision rule, compute the false-alarm rate from the null distribution, and report the resulting detection time for each event at a fixed threshold.
minor comments (5)
  1. [§3.5] The sentence 'm = 10 context steps and n = 5 search steps' is inconsistent with Table 1 and §2.1, where n is the context and m is the search window; please correct the notation.
  2. [Abstract, §4.4] The phrase 'without the need for a labeled training data set of flaring states' is misleading because §4.1 manually labels flaring intervals for exclusion; consider rephrasing to 'without labeled flaring states in the training target' or similar.
  3. [Data Availability] The GitLab URL in the Data Availability section contains a space ('trans finder/blazar flares') and should be replaced with the actual URL.
  4. [Figure 7/8] The VERITAS panels show counts in arbitrary units modified by random noise; the text states that the pipeline uses the true flux, but the figure caption should explicitly state that the displayed counts are a proxy so that readers do not misinterpret the scaling.
  5. [Introduction] The Introduction refers to 'Sect. 3.3' for the simulation study, but the simulation is actually described in §3.4–§3.6; please fix the cross-reference.

Circularity Check

1 steps flagged · score 6.0 of 10

Historical evaluation is in-sample: the training background is defined by excising the same VERITAS flares (plus 50-day buffers) that are then reported as detections, so the headline claim is partly forced by the cleaning rule rather than by learned multiwavelength correlations.

  1. fitted input called prediction [Section 4.1 (historical training cleaning) and Section 4.3/Table 2 (historical results)]
    "We decide to first manually exclude clearly identifiable high states from the light curves. In particular, we remove flaring states based on VERITAS data surrounding MJDs 55,725, 57,200, 57,675, 58,100 and 58,600. We exclude any data taken within 50 days of the respective VHE flares from all channels. ... In general, we find elevated significance during the parts of the light curve that were excluded for training."

    The VHE flares reported as detections in Sections 4.3-4.4 (e.g., t=570, 2,496, 2,900, 3,400 in Table 2) are the same intervals removed when constructing the training background in Section 4.1, each with a 50-day buffer across all channels. All model components (forecaster, autoencoder, BGMM, significance calibration) are fitted on the complement of those intervals, and an anomaly is defined as incompatibility with that cleaned background. The excised intervals are therefore outliers by construction, not out-of-sample predictions. The paper's own statement that elevated significance appears 'during the parts of the light curve that were excluded for training' exposes this coupling. No temporal split, external epoch, or false-alarm budget separates the cleaning rule from the evaluation.

full rationale

The central circularity is the in-sample definition of the training background. In Section 4.1 the paper builds a 'background-only' dataset by manually removing VERITAS flaring states and all data within 50 days of those flares, across all channels, and then applies per-channel variability cutoffs. Section 4.3 then reports high-significance detections at exactly those removed intervals, and Section 4.4 uses these as evidence that the framework 'reliably detect[s] flares' and detects 'flaring states at the same time or earlier than the state-of-the-art.' Because the model and its calibration are trained on the complement of those intervals, the removed intervals are anomalous by construction; the historical detections are therefore a re-detection of the cleaning cuts rather than an independent forecast. The paper acknowledges this coupling only indirectly in the Discussion ('prior data cleaning is needed. This may introduce biases'), but no quantitative control is provided: there is no temporal out-of-sample split, no test on an epoch whose flares were not used to define the background, and no false-alarm budget for the historical comparison. The simulation study is not a valid control for this issue because its background is simulated and its injected flare templates are known, so it cannot calibrate the effect of cleaning real data. The 2020 precursor episode (optical rise detected at t ~ 3896, before the community ATels) is a genuinely more out-of-sample element and partially supports the method's utility, but it does not rescue the main validation of historical VHE flares. The significance-calibration prescription is taken from Sadeh (2020), a coauthor's prior work; this is self-citation, but it is a methodological reference rather than an unverified uniqueness claim, so it is not counted as circular in itself. Overall, the historical evaluation reduces in part to the data-cleaning rule, giving a partial circularity score of 6.

Assumptions & free parameters 10 free parameters · 5 assumptions · 0 invented entities

The central claim rests on modeling choices and data-cleaning assumptions rather than on new physics. The free parameters are mostly hyperparameters of the learned pipeline; the key ones (gamma_signoise, window sizes, cleaning cutoffs) directly set the sensitivity of the anomaly detector. The most consequential assumption is that the cleaned training data are a faithful background model. No new physical entities are introduced.

free parameters (10)
  • gamma_timeaggr = 1 day
    Time bin width defining a single time step; sets the sensitivity timescale to days and is central to the dynamic windowing.
  • gamma_signoise = 5 (standard deviations)
    Threshold used to exclude windows with signal-to-noise above this value from training and calibration; directly controls sensitivity versus false positives.
  • gamma_decay = 1
    Temporal decay rate in the weighting scheme, suppressing old anomalous data in sparse channels.
  • Window sizes n, m = n=10 context steps, m=5 search steps
    Number of steps in the encoder and decoder windows; defines how much history is used and how many future steps are searched.
  • Manual exclusion radius around VHE flares = 50 days
    In Section 4.1, all data within 50 days of each VERITAS flare are removed from training and calibration; this choice directly shapes the background model.
  • Channel cleaning cutoffs = Not quantified (defined per channel variance)
    Section 4.1 excludes data above or below channel-specific flux, count-rate, or magnitude cutoffs; the cutoffs are chosen by hand but not listed numerically.
  • Autoencoder latent dimension = Adaptive
    The embedding dimension is increased until performance stops improving, but the final dimension is not reported.
  • BGMM number of components = Adaptive via MCMC
    The number of mixture components is adaptively increased until a Kolmogorov-Smirnov test shows no improvement; the exact number is not stated.
  • RNN and training hyperparameters = learning rate 0.005, dropout 10%, L2 0.0001; doubled for real data
    Listed in Section 3 and adjusted manually in Section 4.2 to avoid underfitting and overfitting on the same real dataset.
  • Augmentation amplitude ranges = 2 to 100 upward, -2 to -10 downward, duration 1 to 8 days
    Random fluctuations injected during autoencoder training; these ranges are chosen by hand.
assumptions (5)
  • domain assumption The cleaned training windows represent pure background source activity.
    Section 3.1 and Section 4.1 rely on manual and automated cleaning to remove flaring states; the entire anomaly definition is relative to this cleaned background.
  • domain assumption The encoder context windows are free of potential signals during forecasting.
    Section 2 states: 'We train the model under the assumption that the context window is devoid of potential signals, and thus represents a realistic background for flares.'
  • domain assumption The RNN provides unbiased probabilistic forecasts of background activity.
    The residuals tau_res = tau_dec - tau_for assume the forecast is a valid background estimate; biased forecasts would create systematic residuals.
  • domain assumption Measurement uncertainties are Gaussian, independent, and can be propagated by resampling.
    Section 2.1 and Appendix B assume normally distributed uncertainties and sample each time series gamma_err times; the paper notes that nonphysical negative fluxes can arise but are tolerated.
  • domain assumption The null-hypothesis distribution derived from background samples remains valid across the full search grid without explicit trial correction.
    Section 2.5 calibrates test statistics on background examples, but the reported significances are taken as peaks over thousands of time steps and multiple channels; no trial factor is applied.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Early Detection of Multiwavelength Blazar Variability." pith.science (2026). https://pith.science/paper/JYCLYEE7

@misc{pith2026241110140,
  author       = {Pith},
  title        = {Pith review of: Early Detection of Multiwavelength Blazar Variability},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JYCLYEE7}},
  note         = {Machine review of arXiv:2411.10140}
}
read the original abstract

Blazars are a subclass of active galactic nuclei with relativistic jets pointing toward the observer. They are notable for their flux variability at all observed wavelengths and timescales. Together with simultaneous measurements at lower energies, the very-high-energy (VHE) emission observed during blazar flares may be used to probe the population of accelerated particles. However, optimally triggering observations of blazar high states can be challenging. Notable examples include identifying a flaring episode in real time and predicting VHE flaring activity based on lower-energy observables. For this purpose, we have developed a novel deep learning analysis framework, based on data-driven anomaly detection techniques. It is capable of detecting various types of anomalies in real-world, multiwavelength light curves, ranging from clear high states to subtle correlations across bands. Based on unsupervised anomaly detection and clustering methods, we differentiate source variability from noisy background activity, without the need for a labeled training data set of flaring states. The framework incorporates measurement uncertainties and is robust given data quality challenges, such as varying cadences and observational gaps. We evaluate our approach using both historical data and simulations of blazar light curves in two energy bands, corresponding to sources observable with the Fermi Large Area Telescope and the upcoming Cherenkov Telescope Array Observatory. In a statistical analysis, we show that our framework can reliably detect known historical flares.

Figures

Figures reproduced from arXiv: 2411.10140 by the authors.

Figure 1
Figure 1. Illustration of the architecture of the model, as described in the text. 2.1. Light Curves as Input Data The multiwavelength inputs to our pipeline are a set of light curves, obtained from different instruments. Each input (which we refer to in the following as channel) captures an observable related to source brightness (e.g., flux, magnitude), with associated uncertainties. A potential flare is always searched for… view at source ↗
Figure 2
Figure 2. Illustration of a dynamic-window transformation of two light curves into a combined time series structure. The top panel shows two light curve channels, C1 and C2, with a reference time pivot, tref. The context window, τenc, and the search window, τdec, are highlighted as light and dark shaded regions. The middle panel shows a zoom in of both windows over the time relative to tref. The bottom panel shows both window… view at source ↗
Figure 3
Figure 3. Spectral energy distributions for the simulated dataset, inspired by the 2017 steady-state spectrum of the blazar, 1ES 1215+303. Horizontal arrows show the sen￾sitivity regions of Fermi-LAT, VERITAS, and CTAO re￾spectively. The orange dashed (dashed-dotted) line shows the spectrum given in Valverde et al. (2020) for LAT (VERITAS), and the solid green (blue) line shows the com￾bined, EBL-absorbed log-parabola used fo… view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: Left: Bidimensional projections of the embedding space, showing the distributions of background (blue) and flaring states (yellow). 2D projections with PCA on the top left and UMAP on the top right. The color in the PCA and UMAP panels indicates significance (based on …
Figure 5
Figure 5. Figure 5: Left: Significance of flares, captured in three subsequent days, as a function of the simulated flux of the reference source, 1ES 1215+303. The dashed line corresponds to the originally reported flux, Fref, from the 2017 flare of the source (Valverde et al. 2020). Righ…
Figure 6
Figure 6. Figure 6: Illustration of injected flare templates, given scaling factors, 10% (left), 100% (middle) and −15% (right), applied to the baseline flux, Fref. The top panels show light curves in the various channels, as indicated. Energy bins for CTAO (panel rows 1–4) and for Fermi …
Figure 7
Figure 7. Figure 7: Overview of the multiwavelength BL Lac dataset and the corresponding significance of flares, σmm and σcombo. The lowest panel illustrates the contributions of individual channels to the combined significance. Data marked with red crosses are excluded from the training …
Figure 8
Figure 8. Figure 8: Highlighted intervals related to high-significance detections in the BL Lac dataset. (See caption of [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Highlighted interval from the BL Lac dataset. (See caption of [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

61 extracted references · 21 canonical work pages

  1. [1]

    G., Ackermann, M., Adams, J., et al

    Aartsen, M. G., Ackermann, M., Adams, J., et al. 2018, Science, 361, aat1378, doi: 10.1126/science.aat1378

    doi:10.1126/science.aat1378 2018

  2. [2]

    P., Abbott, R., Abbott, T

    Abbott, B. P., Abbott, R., Abbott, T. D., et al. 2017, Phys. Rev. Lett., 119, 161101, doi: 10.1103/PhysRevLett.119.161101

    doi:10.1103/physrevlett.119.161101 2017

  3. [3]

    2022, The Astrophysical Journal Supplement Series, 260, 53, doi: 10.3847/1538-4365/ac6751

    Abdollahi, S., Acero, F., Baldini, L., et al. 2022, The Astrophysical Journal Supplement Series, 260, 53, doi: 10.3847/1538-4365/ac6751

    doi:10.3847/1538-4365/ac6751 2022

  4. [4]

    U., Archambault, S., Archer, A., et al

    Abeysekara, A. U., Archambault, S., Archer, A., et al. 2015, The Astrophysical Journal, 815, L22, doi: 10.1088/2041-8205/815/2/l22

    doi:10.1088/2041-8205/815/2/l22 2015

  5. [5]

    U., Benbow, W., Bird, R., et al

    Abeysekara, A. U., Benbow, W., Bird, R., et al. 2018, ApJ, 856, 95, doi: 10.3847/1538-4357/aab35c

    doi:10.3847/1538-4357/aab35c 2018

  6. [6]

    A., Aliu, E., Arlen, T., et al

    Acciari, V. A., Aliu, E., Arlen, T., et al. 2009, The Astrophysical Journal, 693, L104, doi: 10.1088/0004-637x/693/2/l104

    work page doi:10.1088/0004-637x/693/2/l104 2009

  7. [7]

    2013, Astroparticle Physics, 43, 3, doi: 10.1016/j.astropartphys.2013.01.007

    Acharya, B., Actis, M., Aghajani, T., et al. 2013, Astroparticle Physics, 43, 3, doi: 10.1016/j.astropartphys.2013.01.007

    doi:10.1016/j.astropartphys.2013.01.007 2013

  8. [8]

    2022, Science, 376, abn0567, doi: 10.1126/science.abn0567

    Aharonian, F., Ait Benkhali, F., Ang¨ uner, E., et al. 2022, Science, 376, abn0567, doi: 10.1126/science.abn0567

    doi:10.1126/science.abn0567 2022

Show all 61 references
  1. [9]

    G., Bazer-Bachi, A

    Aharonian, F., Akhperjanian, A. G., Bazer-Bachi, A. R., et al. 2007, The Astrophysical Journal, 664, L71, doi: 10.1086/520635

    2007 doi

  2. [10]

    2006, Astron

    Aharonian, F., et al. 2006, Astron. Astrophys., 457, 899, doi: 10.1051/0004-6361:20065351

    2006 doi

  3. [11]

    2008, Physics Letters B, 668, 253, doi: 10.1016/j.physletb.2008.08.053 Aleksi´ c, J., et al

    Albert, J., Aliu, E., Anderhub, H., et al. 2008, Physics Letters B, 668, 253, doi: 10.1016/j.physletb.2008.08.053 Aleksi´ c, J., et al. 2016, Astroparticle Physics, 72, 76, doi: 10.1016/j.astropartphys.2015.02.005

    2008 doi

  4. [12]

    2013, ApJ, 762, 92, doi: 10.1088/0004-637X/762/2/92

    Arlen, T., Aune, T., Beilicke, M., et al. 2013, ApJ, 762, 92, doi: 10.1088/0004-637X/762/2/92

    2013 doi

  5. [13]

    B., Abdo, A

    Atwood, W. B., Abdo, A. A., Ackermann, M., et al. 2009, ApJ, 697, 1071, doi: 10.1088/0004-637X/697/2/1071

    2009 doi

  6. [14]

    1999, A&AS, 135, 371, doi: 10.1051/aas:1999179

    Ballet, J. 1999, A&AS, 135, 371, doi: 10.1051/aas:1999179

    1999 doi

  7. [15]

    C., Kulkarni, S

    Bellm, E. C., Kulkarni, S. R., Graham, M. J., et al. 2019a, PASP, 131, 018002, doi: 10.1088/1538-3873/aaecbe

    doi

  8. [16]

    C., Kulkarni, S

    Bellm, E. C., Kulkarni, S. R., Barlow, T., et al. 2019b, PASP, 131, 068003, doi: 10.1088/1538-3873/ab0c2a

    doi

  9. [17]

    2022, Galaxies, 10, doi: 10.3390/galaxies10020039

    Biteau, J., & Meyer, M. 2022, Galaxies, 10, doi: 10.3390/galaxies10020039

    2022 doi

  10. [18]

    M., & Jordan, M

    Blei, D. M., & Jordan, M. I. 2006, Bayesian Analysis, 1, 121 , doi: 10.1214/06-BA104

    2006 doi

  11. [19]

    N., Hill, J

    Burrows, D. N., Hill, J. E., Nousek, J. A., et al. 2005, SSRv, 120, 165, doi: 10.1007/s11214-005-5097-2

    2005 doi

  12. [20]

    2020, Galaxies, 8, doi: 10.3390/galaxies8040072 Cherenkov Telescope Array Consortium, Acharya, B

    Cerruti, M. 2020, Galaxies, 8, doi: 10.3390/galaxies8040072 Cherenkov Telescope Array Consortium, Acharya, B. S.,

    2020 doi

  13. [21]

    2019, Science with the Cherenkov Telescope Array, doi: 10.1142/10986

    Agudo, I., et al. 2019, Science with the Cherenkov Telescope Array, doi: 10.1142/10986

    2019 doi

  14. [22]

    Cheung, C. C. 2020, The Astronomer’s Telegram, 13933, 1 D’Ammando, F. 2020a, The Astronomer’s Telegram, 14065, 1 —. 2020b, The Astronomer’s Telegram, 14069, 1

    2020

  15. [23]

    2017, in International Cosmic Ray Conference, Vol

    Deil, C., Zanin, R., Lefaucheur, J., et al. 2017, in International Cosmic Ray Conference, Vol. 301, 35th International Cosmic Ray Conference (ICRC2017), 766, doi: 10.48550/arxiv.1709.01751

    2017 doi arXiv

  16. [24]

    A., Beardmore, A

    Evans, P. A., Beardmore, A. P., Page, K. L., et al. 2007, A&A, 469, 379, doi: 10.1051/0004-6361:20077530

    2007 doi

  17. [25]

    Feng, Q., & Lin, T. T. Y. 2016, Proceedings of the International Astronomical Union, 12, 173–179, doi: 10.1017/S1743921316012734

    2016 doi

  18. [26]

    G., et al

    Feng, Q., VERITAS Collaboration, Jorstad, S. G., et al. 2017, in International Cosmic Ray Conference, Vol. 301, 35th International Cosmic Ray Conference (ICRC2017), 648, doi: 10.22323/1.301.0648

    2017 doi

  19. [27]

    D., Razzaque, S., & Dermer, C

    Finke, J. D., Razzaque, S., & Dermer, C. D. 2010, ApJ, 712, 238, doi: 10.1088/0004-637X/712/1/238 F.R.S., K. P. 1901, The London, Edinburgh, and Dublin Philosophical Magazine and Journal of Science, 2, 559, doi: 10.1080/14786440109462720

    2010 doi

  20. [28]

    2019, The Astronomer’s Telegram, 12718, 1

    Garrappa, S., & Buson, S. 2019, The Astronomer’s Telegram, 12718, 1

    2019

  21. [29]

    2004, ApJ, 611, 1005, doi: 10.1086/422091 25

    Gehrels, N., Chincarini, G., Giommi, P., et al. 2004, ApJ, 611, 1005, doi: 10.1086/422091 25

    2004 doi

  22. [30]

    S., & Larionov, V

    Grishina, T. S., & Larionov, V. M. 2020, The Astronomer’s Telegram, 13930, 1

    2020

  23. [31]

    A., Peck, A

    Gurwell, M. A., Peck, A. B., Hostler, S. R., Darrah, M. R., & Katz, C. A. 2007, in Astronomical Society of the Pacific Conference Series, Vol. 375, From Z-Machines to ALMA: (Sub)Millimeter Spectroscopy of Galaxies, ed. A. J. Baker, J. Glenn, A. I. Harris, J. G. Mangum, & M. S....

    2007 doi

  24. [32]

    2008, AIP Conference Proceedings, 1085, 657, doi: 10.1063/1.3076760 IRSA

    Holder, J., et al. 2008, AIP Conference Proceedings, 1085, 657, doi: 10.1063/1.3076760 IRSA. 2022, Time Series Tool, IPAC, doi: 10.26131/IRSA538

    2008 doi

  25. [33]

    2020, The Astronomer’s Telegram, 13956, 1

    Jankowsky, F., & Wagner, S. 2020, The Astronomer’s Telegram, 13956, 1

    2020

  26. [34]

    P., & Ba, J

    Kingma, D. P., & Ba, J. 2014, arXiv preprint arXiv:1412.6980

    2014 arXiv

  27. [35]

    2000, The Astrophysical Journal, 542, 235, doi: 10.1086/309533

    Achterberg, A. 2000, The Astrophysical Journal, 542, 235, doi: 10.1086/309533

    2000 doi

  28. [36]

    M., Kulkarni, S

    Law, N. M., Kulkarni, S. R., Dekany, R. G., et al. 2009, PASP, 121, 1395, doi: 10.1086/648598 MAGIC Collaboration, Acciari, V. A., et al. 2019, Nature, 575, 455, doi: 10.1038/s41586-019-1750-x

    2009 doi

  29. [37]

    2017, in Proceedings, 35th International Cosmic Ray Conference (ICRC2017):

    Maier, G., & Holder, J. 2017, in Proceedings, 35th International Cosmic Ray Conference (ICRC2017):

    2017

  30. [38]

    L., Pruzhinskaya, M

    Malanchev, K. L., Pruzhinskaya, M. V., Korolev, V. S., et al. 2021, MNRAS, 502, 5147, doi: 10.1093/mnras/stab316

    2021 doi

  31. [39]

    2018, arXiv preprint arXiv:1802.03426

    McInnes, L., Healy, J., & Melville, J. 2018, arXiv preprint arXiv:1802.03426

    2018 arXiv

  32. [40]

    J., & Peel, D

    McLachlan, G. J., & Peel, D. 2000, Finite mixture models, Vol. 299 (John Wiley & Sons)

    2000

  33. [41]

    2020, The Astronomer’s Telegram, 14072, 1

    Mereu, I. 2020, The Astronomer’s Telegram, 14072, 1

    2020

  34. [42]

    2018, IEEE Access, 6, 39501

    Min, E., Guo, X., Liu, Q., et al. 2018, IEEE Access, 6, 39501

    2018

  35. [43]

    2019, The Astronomer’s Telegram, 12724, 1

    Mirzoyan, R. 2019, The Astronomer’s Telegram, 12724, 1

    2019

  36. [44]

    2016, The Astronomer’s Telegram, 9599, 1

    Mukherjee, R., & VERITAS Collaboration. 2016, The Astronomer’s Telegram, 9599, 1

    2016

  37. [45]

    D., Takami, H., & Migliori, G

    Murase, K., Dermer, C. D., Takami, H., & Migliori, G. 2012, The Astrophysical Journal, 749, 63, doi: 10.1088/0004-637x/749/1/63 Nieto Casta˜ no, D., Brill, A., Kim, B., & Humensky, T. B. 2017, PoS, ICRC2017, 809, doi: 10.22323/1.301.0809

    2012 doi

  38. [46]

    O., et al

    Nilsson, K., Lindfors, E., Takalo, L. O., et al. 2018, A&A, 620, A185, doi: 10.1051/0004-6361/201833621

    2018 doi

  39. [47]

    2019, A&A, 631, A147, doi: 10.1051/0004-6361/201935634

    Nordin, J., Brinnel, V., van Santen, J., et al. 2019, A&A, 631, A147, doi: 10.1051/0004-6361/201935634

    2019 doi

  40. [48]

    1996, The VizieR database of astronomical catalogues, CDS, Centre de Donn ˜A©es astronomiques de Strasbourg, doi: 10.26093/CDS/VIZIER

    Ochsenbein, F. 1996, The VizieR database of astronomical catalogues, CDS, Centre de Donn ˜A©es astronomiques de Strasbourg, doi: 10.26093/CDS/VIZIER

    1996 doi

  41. [49]

    2000, A&AS, 143, 23, doi: 10.1051/aas:2000169

    Ochsenbein, F., Bauer, P., & Marcout, J. 2000, A&AS, 143, 23, doi: 10.1051/aas:2000169

    2000 doi

  42. [50]

    2020, The Astronomer’s Telegram, 13964, 1

    Ojha, R., & Valverde, J. 2020, The Astronomer’s Telegram, 13964, 1

    2020

  43. [51]

    Ong, R. A. 2011, The Astronomer’s Telegram, 3459, 1

    2011

  44. [52]

    2015, in Proceedings, 34th International Cosmic Ray Conference (ICRC2015): The Hague, The

    Park, N. 2015, in Proceedings, 34th International Cosmic Ray Conference (ICRC2015): The Hague, The

    2015

  45. [53]

    2018, MNRAS, 476, 2117, doi: 10.1093/mnras/sty348

    Reis, I., Poznanski, D., Baron, D., Zasowski, G., & Shahaf, S. 2018, MNRAS, 476, 2117, doi: 10.1093/mnras/sty348

    2018 doi

  46. [54]

    2018, 854, 54, doi: 10.3847/1538-4357/aaa7ee

    Winter, W. 2018, 854, 54, doi: 10.3847/1538-4357/aaa7ee

    2018 doi

  47. [55]

    2020, ApJL, 894, L25, doi: 10.3847/2041-8213/ab8b5f

    Sadeh, I. 2020, ApJL, 894, L25, doi: 10.3847/2041-8213/ab8b5f

    2020 doi

  48. [56]

    B., & Lahav, O

    Sadeh, I., Abdalla, F. B., & Lahav, O. 2016, PASP, 128, 104502, doi: 10.1088/1538-3873/128/968/104502

    2016 doi

  49. [57]

    2015, Monthly Notices of the Royal Astronomical Society, 450, 183, doi: 10.1093/mnras/stv641

    Sironi, L., Petropoulou, M., & Giannios, D. 2015, Monthly Notices of the Royal Astronomical Society, 450, 183, doi: 10.1093/mnras/stv641

    2015 doi

  50. [58]

    2020, The Astronomer’s Telegram, 13958, 1

    Steineke, R., Waller, L., Reinhart, D., et al. 2020, The Astronomer’s Telegram, 13958, 1

    2020

  51. [59]

    2020, The Astrophysical Journal, 891, 170, doi: 10.3847/1538-4357/ab765d

    Valverde, J., Horan, D., Bernard, D., et al. 2020, The Astrophysical Journal, 891, 170, doi: 10.3847/1538-4357/ab765d

    2020 doi

  52. [60]

    2017, in International Cosmic Ray Conference, Vol

    Wood, M., Caputo, R., Charles, E., et al. 2017, in International Cosmic Ray Conference, Vol. 301, 35th International Cosmic Ray Conference (ICRC2017), 824, doi: 10.48550/arxiv.1707.09551

    2017 doi arXiv

  53. [61]

    2024, MNRAS, 534, 2142, doi: 10.1093/mnras/stae2082

    Siemiginowska, A. 2024, MNRAS, 534, 2142, doi: 10.1093/mnras/stae2082

    2024 doi

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.